Subcutaneously injected insulin has been used to treat diabetes mellitus since the 1920s. Only recently, however, has the long-term benefit of tight glycemic control with aggressive insulin therapy been demonstrated. Despite studies showing its benefits, aggressive insulin therapy has been slow to gain acceptance in clinical practice settings. One limitation is the inconvenience and poor subject acceptability of injection therapy, especially a multiple daily injection regimen. Hence, clinical realization of the benefits of aggressive insulin therapy has been limited by shortcomings of available subcutaneous (SC) delivery methods. The present study investigates the clinical utility of pre-meal pulmonary delivery of insulin in the therapy of subjects with diabetes mellitus. To maximize the efficiency and reproducibility of pulmonary insulin delivery, a novel dry powder insulin formulation and aerosol delivery device have been developed. This aerosol delivery system is designed to permit safe, noninvasive delivery of rapid-acting insulin in 1-2 inhalations. The approximate dosage bioequivalence of inhaled human insulin relative to SC Regular insulin is 1 mg inhaled insulin about 3 U SC insulin. Three Phase II outpatient studies have been conducted, involving a total of approximately 180 subjects with type 1 or type 2 diabetes mellitus. These were 3-month treatment trials that randomized half the subjects to inhaled insulin and half to a continuation of usual therapy; after the 3-month treatment period, subjects could choose to participate in a nonrandomized extension trial of inhaled insulin therapy. Two of these studies demonstrated an inhaled insulin regimen to be comparably efficacious to a conventional subcutaneously injected insulin regimen. The third study showed that in type 2 subjects who were not well controlled on oral agents, addition of inhaled insulin to their regimen markedly improved glycemic control. Inhaled insulin has been very well tolerated in these studies and has had no effect on pulmonary function tests. It has also been very well liked by subjects, the great majority of whom (>80%) have opted to continue in the extension trials on inhaled insulin.